The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from multiple external sources which could contain hidden malicious instructions designed to influence the architectural blueprint or subsequent agent tasks.
Ingestion points: The skill reads all files within design/gdd/ and docs/engine-reference/, including user-provided game concept and systems index documents.
Boundary markers: The instructions do not define any delimiters (e.g., XML tags or triple quotes) or specific "ignore embedded instructions" warnings when interpolating the extracted technical requirements into the architecture document or when presenting them to the user.
Capability inventory: The skill has the ability to Write to the filesystem (docs/architecture/architecture.md), spawn new agent tasks via the Task tool (e.g., lead-programmer), and execute Bash commands.
Sanitization: There is no evidence of sanitization, escaping, or validation logic to ensure that extracted requirements do not contain executable commands or prompt injection payloads.

create-architecture